Eavesdropping is predicted to evolve between sympatric, but not allopatric, predator and prey. The evolutionary arms race between Asian honey bees and their hornet predators has led to a remarkable defence, heat balling, which suffocates hornets with heat and carbon dioxide. We show that the sympatric Asian species, Apis cerana (Ac), formed heat balls in response to Ac and hornet (Vespa velutina) alarm pheromones, demonstrating eavesdropping. The allopatric species, Apis mellifera (Am), only weakly responded to a live hornet and Am alarm pheromone, but not to hornet alarm pheromone. We observed typical hornet alarm pheromone-releasing behaviour, hornet sting extension, when guard bees initially attacked. Once heat balls were formed, guards released honey bee sting alarm pheromones: isopentyl acetate, octyl acetate, (E)-2-decen-1-yl acetate and benzyl acetate. Only Ac heat balled in response to realistic bee alarm pheromone component levels (<1 bee-equivalent, 1 mg) of isopentyl acetate. Detailed eavesdropping experiments showed that Ac, but not Am, formed heat balls in response to a synthetic blend of hornet alarm pheromone. Only Ac antennae showed strong, consistent responses to hornet alarm pheromone compounds and venom volatiles. These data provide the first evidence that the sympatric Ac, but not the allopatric Am, can eavesdrop upon hornet alarm pheromone and uses this information, in addition to bee alarm pheromone, to heat ball hornets. Evolution has likely given Ac this eavesdropping ability, an adaptation that the allopatric Am does not possess.

Referential communication provides a sophisticated way in which animals can communicate information about their environment. Previously, research demonstrated that honey bee stop signals encode predator danger in their fundamental frequency and danger context in their duration. Here, we show that these signals also encode danger in their vibrational amplitude. Stop signals elicited by the more dangerous predator, the large hornet (Vespa mandarinia) had significantly 1.5-fold higher vibrational amplitudes than those elicited by the small hornet predator (Vespa velutina). We measured the freezing vibrational response thresholds, and show that natural signals exceed these response thresholds. Finally, with artificial playbacks of the vibratory stop signal, we demonstrate that these signals referentially encode the danger that foragers experience at food source. Stop signals elicited by the larger and significantly more dangerous predator (V. mandarinia) were significantly 1.4-fold more inhibitory than stop signals elicited by the smaller and less dangerous predator (V. velutina).

Resisting majesty: Apis cerana, has lower antennal sensitivity and decreased attraction to queen mandibular pheromone than Apis mellifera In highly social bees, queen mandibular pheromone (QMP) is vital for colony life. Both Apis cerana (Ac) and Apis mellifera (Am) share an evolutionarily conserved set of QMP compounds: (E)-9-oxodec-2-enoic acid (9-ODA), (E)-9-hydroxydec-2-enoic acid (9-HDA), (E)-10-hydroxy-dec-2-enoic acid (10-HDA), 10-hydroxy-decanoic acid (10-HDAA), and methyl p-hydroxybenzoate (HOB) found at similar levels. However, evidence suggests there may be species-specific sensitivity differences to QMP compounds because Ac workers have higher levels of ovarian activation than Am workers. Using electroantennograms, we found species-specific sensitivity differences for a blend of the major QMP compounds and three individual compounds (9-HDA, 10-HDAA, and 10-HDA). As predicted, Am was more sensitive than Ac in all cases (1.3-to 2.7-fold higher responses). There were also species differences in worker retinue attraction to three compounds (9-HDA, HOB, and 10-HDA). In all significantly different cases, Am workers were 4.5-to 6.2-fold more strongly attracted than Ac workers were. Thus, Ac workers responded less strongly to QMP than Am workers, and 9-HDA and 10-HDA consistently elicited stronger antennal and retinue formation responses.Honey bee queens produce a pheromone, queen mandibular pheromone (QMP), which plays a central role in colony life and has multiple effects, depending upon the receivers and the context [1][2][3] . QMP can act as a sex pheromone and attract drones to virgin queens 3,4 . Within the colony, QMP signals the queen's presence, inhibits worker ovarian development 5,6 , and maintains normal colony activity 7 . Interestingly, workers of the Asian honey bee, Apis cerana (Ac), have higher rates of ovarian activation than workers of the other Apis species, including A. mellifera ligustica (Am) and A. florea (Af) 8,9 . In colonies with a normal egg-laying queen (queenright colonies), about 5% of Ac workers have activated ovaries [10][11][12] . In comparison, 0.02% of Am workers and 0.01% of Af workers have activated ovaries 10,13 . Sakagami and Akahra (1958) similarly reported that about 10-20% of Ac workers contained mature eggs in their ovaries 14 . In contrast, about one Am worker in 1,000 contains visible eggs and only one worker in 10,000 contains a full-sized egg 8 . QMP is also essential for creating the worker cluster (retinue) around the queen 3,10,13 . The attraction exerted by QMP reflects its central role and the importance of this retinue for grooming and feeding the queen and distributing QMP throughout the colony 7 . Aside from daily care, this retinue has implications for queen survival. For example, Am workers are more attracted to higher-as compared to lower-quality queens, and low queen attractiveness may contribute to queen replacement, a process in which workers play can play a role 15 . Most, QMP retinue studies have focused on Am 7 , but QMP also elicits retinue at...

Mating is an important process in bumblebees that could affect queen diapause survival and offspring reproduction. Both queens and males could influence mating failure. Here, we used the indigenous bumblebee Bombus lantschouensis to evaluate the mating interactions of males and gynes. The effects of kin recognition and males and gynes from multiple colonies on mating latency, mating duration, and the mating success rate were investigated. The results showed that gynes mated with related males had a longer mating latency than those mated with unrelated males (42.88 ± 3.8 min and 24.15 ± 2.5 min, respectively, P < 0.05) and that the mating success rate was significantly higher in unrelated groups than in related groups (49.29 ± 4.1% and 36.74 ± 2.6%, respectively, P < 0.05). However, no preference for related or unrelated males was observed in the mixed mating groups (mating success rate 30.63 ± 3.1% and 30.73 ± 2.4%, respectively, P > 0.05). Interestingly, the occurrence of males from multiple sources significantly increased the mating success rate (one colony 39.1 ± 4.2% to four colonies 60.05 ± 5.7%, P < 0.05). Nonetheless, an increase in the number of gyne sources had no effect on the mating success rate (one colony 39.50 ± 4.9% to four colonies 43.52 ± 5.7%, P > 0.05). Mating latency was significantly more influenced by males and gynes from multiple colonies than by kin relationship, male number, and gyne number (P < 0.05). In conclusion, there is no evidence that the bumblebee B. lantschouensis can recognize kin relationships before mating. The presence of males and gynes from multiple colonies can influence mating latency. Moreover, males from multiple colonies can significantly enhance mating success, which has implications for bumblebee ecological conservation and artificial mass rearing.